ABSTRACT There are known anatomic and physiologic risk factors for sleep apnea (OSA). One strategy to discover how these factors together contribute to OSA risk is to examine patients with clinically defined extreme phenotypes, i.e., patients whose disease status is not explained by primary clinical risk factors. As obesity is the primary risk factor for OSA, two such phenotypes are OSA patients with BMI?25 kg/m2 and non-apneic controls with BMI?35 kg/m2. Apneics with BMI?25 have OSA despite not having the obesity risk factor. Non- apneic controls with BMI?35 are protected from OSA despite high obesity levels; the mechanisms protecting these individuals are unknown. Thus, the overall objectives of this proposal are directed towards understanding the anatomical and physiological factors determining these two extreme OSA phenotypes. Most individuals with a BMI?35 have OSA and most individuals with a BMI?25 do not. The primary question is, ?What protects individuals with a BMI?35 from OSA and why do some individuals with a BMI?25 have OSA?? Based on our preliminary data, the global hypotheses motivating this study are that: 1) in obese controls (BMI?35) tongue fat and parapharyngeal fat pad volumes will be reduced, pharyngeal muscle responsiveness will be enhanced and airway collapsibility will be less compromised compared to obese apneics, protecting these obese subjects from OSA (Aim 1); 2) in thin apneics (BMI?25) mandibular depth (a measure of retrognathia) will be smaller, UA total soft tissue volume will be larger, loop gain will be higher, and airway collapsibility will be more compromised compared to non-apneics, causing OSA in these thin subjects (Aim 2); and 3) using machine learning techniques, we will assess complex interactions between all anatomic and physiologic traits that predict OSA status in each extreme phenotype (Aim 3). We have the following Specific Aims: Specific Aim 1: To quantitatively assess upper airway anatomic phenotypes with magnetic resonance imaging (MRI) and physiologic phenotypes in gender, age, BMI and race matched apneics and non- apneic controls with a BMI ? 35 kg/m2. Specific Aim 2: To quantitatively assess UA anatomic phenotypes with MRI and physiologic phenotypes in gender, age, BMI and race matched apneics and controls with a BMI ? 25 kg/m2. Specific Aim 3: Utilize machine learning techniques (CART and Random Forest) to explore pathways to OSA using the anatomic and physiologic domains and their interactions, in both extreme phenotypes. This analysis will identify the most important combinations of anatomic and physiologic risk/protective factors for OSA, serving to both validate the hypotheses in Aims 1-2 and identify novel disease pathways. This proposal has multiple innovative features, but the unique innovation is the examination of two extreme OSA phenotypes. The project will lead to better understanding of the relative contribution of each anatomic/physiologic trait to the development of OSA within a given individual, which may allow targeting appropriate trait(s) with specific and novel treatments. This is a primary goal of P4 medicine, an overarching theme to the overall Program Project.